def setUpClass(cls):
super().setUpClass()
cls.agent = SAC(state_shape=cls.continuous_env.observation_space.shape,
action_dim=cls.continuous_env.action_space.low.size,
batch_size=cls.batch_size,
auto_alpha=True,
gpu=-1)
def main():
""" Main function to be run. """
parser = argparse.ArgumentParser(description='Test force control')
parser.add_argument('poldir',
type=str,
default=None,
help='Policy directory')
parser.add_argument('-c',
'--common-test',
action='store_true',
help='Use common test parameters ')
parser.add_argument('-n', type=int, help='number of tests', default=1)
parser.add_argument('--custom-path',
action='store_true',
help='Execute custom path for successful tests')
parser.add_argument('--training',
action='store_true',
help='Execute actions within random std')
args = parser.parse_args()
rospy.init_node('ur3e_tf2rl', anonymous=True, log_level=rospy.ERROR)
clear_gym_params('ur_gym')
policy_dir = os.path.join(os.getcwd(), args.poldir)
assert os.path.exists(policy_dir)
env = load_env(policy_dir, args.common_test)
actor_class = rospy.get_param("ur_gym/actor_class", "default")
policy = SAC(
state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
max_action=env.action_space.high[0],
actor_class=actor_class,
)
rospy.set_param('ur_gym/output_dir', policy_dir)
checkpoint = tf.train.Checkpoint(policy=policy)
_latest_path_ckpt = tf.train.latest_checkpoint(policy_dir)
checkpoint.restore(_latest_path_ckpt).expect_partial()
print("Restored {}".format(_latest_path_ckpt))
test_policy(env, policy, args.n, args.custom_path, args.training)
parser.set_defaults(batch_size=100)
parser.set_defaults(n_warmup=0)
parser.set_defaults(max_steps=5e4)
parser.set_defaults(episode_max_steps=150)
parser.set_defaults(model_dir=args.model_path + args.model_name)
args = parser.parse_args()
env = gym.make(args.env_name)
test_env = gym.make(args.env_name)
policy_ = SAC(state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
gpu=-1,
memory_capacity=args.memory_capacity,
max_action=env.action_space.high,
batch_size=args.batch_size,
n_warmup=args.n_warmup,
alpha=args.alpha,
auto_alpha=args.auto_alpha,
lr=1e-5)
trainer = Trainer(policy_, env, args, test_env=test_env)
#trainer()
current_steps = 0
max_steps = 5
total_steps = 0
episode_max_steps = 10
while total_steps <= max_steps:
import roboschool
import gym
from tf2rl.algos.sac import SAC
from tf2rl.experiments.trainer import Trainer
if __name__ == '__main__':
parser = Trainer.get_argument()
parser = SAC.get_argument(parser)
parser.add_argument('--env-name', type=str, default="RoboschoolAnt-v1")
parser.set_defaults(batch_size=100)
parser.set_defaults(n_warmup=10000)
args = parser.parse_args()
env = gym.make(args.env_name)
test_env = gym.make(args.env_name)
policy = SAC(
state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
gpu=args.gpu,
memory_capacity=args.memory_capacity,
max_action=env.action_space.high[0],
batch_size=args.batch_size,
n_warmup=args.n_warmup)
trainer = Trainer(policy, env, args, test_env=test_env)
trainer()
import roboschool, gym
from tf2rl.algos.sac import SAC
from tf2rl.trainer.trainer import Trainer
if __name__ == '__main__':
parser = Trainer.get_argument()
parser.add_argument('--env-name', type=str, default="RoboschoolAnt-v1")
args = parser.parse_args()
env = gym.make(args.env_name)
test_env = gym.make(args.env_name)
policy = SAC(state_dim=env.observation_space.high.size,
action_dim=env.action_space.high.size,
gpu=args.gpu,
batch_size=100)
trainer = Trainer(policy, env, args, test_env=test_env)
trainer()
def run(parser):
args = parser.parse_args()
if args.gpu < 0:
tf.config.experimental.set_visible_devices([], 'GPU')
else:
physical_devices = tf.config.list_physical_devices('GPU')
tf.config.set_visible_devices(physical_devices[args.gpu], 'GPU')
tf.config.experimental.set_virtual_device_configuration(
physical_devices[args.gpu], [
tf.config.experimental.VirtualDeviceConfiguration(
memory_limit=1024 * 3)
])
if args.env == 200:
envname = 'ScratchItchPR2X'
elif args.env == 201:
envname = 'DressingPR2X'
elif args.env == 202:
envname = 'BedBathingPR2X'
logdir = f'MFBox_Assistive'
if args.SAC:
wandb.init(config=vars(args),
project="Assistive Gym",
name=f'SAC on {envname}')
elif args.PPO:
wandb.init(config=vars(args),
project="Assistive Gym",
name=f'PPO on {envname}')
elif args.TD3:
wandb.init(config=vars(args),
project="Assistive Gym",
name=f'TD3 on {envname}')
elif args.DEBUG:
logdir = f'DEBUG_Assistive'
wandb.init(config=vars(args),
project="Assistive Gym",
name=f'DEBUG on {envname}')
else:
print('PLEASE INDICATE THE ALGORITHM !!')
if not os.path.exists(logdir):
os.makedirs(logdir)
parser.set_defaults(logdir=logdir)
args = parser.parse_args()
env = gym.make(f'{envname}-v0')
#test_env = Monitor(env,logdir,force=True)
test_env = gym.make(f'{envname}-v0')
if args.SAC:
policy = SAC(state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
gpu=args.gpu,
memory_capacity=args.memory_capacity,
max_action=env.action_space.high[0],
batch_size=args.batch_size,
n_warmup=args.n_warmup,
alpha=args.alpha,
auto_alpha=args.auto_alpha)
trainer = Trainer(policy, env, args, test_env=test_env)
elif args.PPO:
policy = PPO(state_shape=env.observation_space.shape,
action_dim=get_act_dim(env.action_space),
is_discrete=is_discrete(env.action_space),
max_action=None if is_discrete(env.action_space) else
env.action_space.high[0],
batch_size=args.batch_size,
actor_units=(64, 64),
critic_units=(64, 64),
n_epoch=10,
lr_actor=3e-4,
lr_critic=3e-4,
hidden_activation_actor="tanh",
hidden_activation_critic="tanh",
discount=0.99,
lam=0.95,
entropy_coef=0.,
horizon=args.horizon,
normalize_adv=args.normalize_adv,
enable_gae=args.enable_gae,
gpu=args.gpu)
trainer = OnPolicyTrainer(policy, env, args, test_env=test_env)
elif args.TD3:
policy = TD3(state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
gpu=args.gpu,
memory_capacity=args.memory_capacity,
max_action=env.action_space.high[0],
batch_size=args.batch_size,
n_warmup=args.n_warmup)
trainer = Trainer(policy, env, args, test_env=test_env)
elif args.DEBUG:
policy = SAC(state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
gpu=args.gpu,
memory_capacity=args.memory_capacity,
max_action=env.action_space.high[0],
batch_size=args.batch_size,
n_warmup=100,
alpha=args.alpha,
auto_alpha=args.auto_alpha)
parser.set_defaults(test_interval=200)
args = parser.parse_args()
trainer = Trainer(policy, env, args, test_env=None)
trainer()
ros_param_path = load_ros_params(rospackage_name="ur_rl",
rel_path_from_package_to_file="config",
yaml_file_name=param_file)
args.episode_max_steps = rospy.get_param("ur_gym/steps_per_episode", 200)
env = load_environment(
rospy.get_param('ur_gym/env_id'),
max_episode_steps=args.episode_max_steps)
actor_class = rospy.get_param("ur_gym/actor_class", "default")
policy = SAC(
state_shape=env.observation_space.shape,
action_dim=env.action_space.high.size,
max_action=env.action_space.high[0],
actor_class=actor_class,
batch_size=args.batch_size,
n_warmup=args.n_warmup,
auto_alpha=args.auto_alpha,
lr=args.lr
)
trainer = Trainer(policy, env, args, test_env=None)
outdir = trainer._output_dir
rospy.set_param('ur_gym/output_dir', outdir)
log_ros_params(outdir)
copyfile(ros_param_path, outdir + "/ros_gym_env_params.yaml")
trainer()
print("duration", (timeit.default_timer() - start_time)/60.,"min")
# rosrun ur3e_rl tf2rl_sac.py --env-id=0
